MySQL NDB Cluster processes like ndb_mgmd and ndbmtd are now discovered by Agents when the Agents are installed upon the same host.

When ndb_mgmd and ndbmtd processes are discovered, Agents look for the --ndb-connectstring and --ndb-nodeid parameters from the process arguments. They use this information to connect to the management node and do a full discovery of the NDB Cluster node configuration.

This information is transmitted to the Service Manager and become selectable in the User Interface much like Replication Groups.

When selected, all of their individual node types such as NDB Data Nodes, Management Nodes, and API nodes are also able to be directly targeted. This information is augmented with ndbinfo data collected from the SQL nodes, including from the new ndbinfo.processes table added in MySQL NDB Cluster 7.5. All of this is combined to model and display the full topology and state of all processes running within an NDB Cluster.

The Topology page has been extended to show NDB Cluster state quickly and easily. Management, Data and API nodes are nested and grouped together visually, and show the computed state at each and every layer. The new NDB Cluster Status Advisor monitors the entire NDB Cluster health using this data. Should a single NDB Data Node fail within a Node Group, a Critical event will be generated and the topology will be shown as DEGRADED. If all nodes in a Data Node Group fail, an Emergency event will be generated and the topology will show as OFFLINE.

The new NDB Cluster Memory report breaks down the output of the ndbinfo.memory_per_fragment table, giving you greater insight in to how NDB Data Memory is getting utilized across the Cluster. The report starts from a Schema-centric view and allows you to drill into how objects balance Data Memory across nodes within the NDB Cluster.

Drilling into specific objects, you can easily see how memory is used per fragment and on which NDB Data Nodes they reside. By default the report shows usage by total memory allocated. You can use the menu to choose other detailed metrics such as Fixed Size or Variable Size Element allocations.